摘要
为提高舰船磁隐身能力,对舰船涡流磁场的深度换算进行研究,以满足消磁勤务的要求.首先,采用COMSOL仿真软件,使用有限元法进行铜板涡流磁场建模,通过外加磁场的改变模拟产生的两个深度涡流磁场,建立涡流磁场数据库;然后,在此基础上使用条件生成对抗神经网络(CGAN)对涡流磁场数据进行训练和学习,建立不同深度的涡流磁场推算模型.得到的模型对涡流磁场的推算准确度较高,训练结果较好,对实际的涡流磁场研究有指导意义.
In order to improve the magnetic stealth capability of ships, the depth conversion of eddy current magnetic field was studied to meet the requirements of demagnetization service.First of all,the COMSOL simulation software was adopted to use the finite element method to model the eddy current magnetic field of the copper plate. The two deep eddy current magnetic fields generated by the change of the applied magnetic field were simulated to establish the eddy current magnetic field database.Then,the conditional generative adversarial neural network(CGAN) was used on this basis,to train and learn the eddy current magnetic field data. Estimation models of eddy current magnetic fields of different depths were established. The obtained model has high accuracy in calculating the eddy current magnetic field, and the training result is good, which has guiding significance for the actual eddy current magnetic field research.
作者
刘胜道
文昊东
何保委
欧阳剑锋
LIU Shengdao;WEN Haodong;HE Baowei;OUYANG Jianfeng(School of Electronic Engineering,Naval University of Engineering,Wuhan 430033,China)
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2021年第8期116-120,共5页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(51377165)。
关键词
涡流磁场
条件生成对抗神经网络(CGAN)
深度换算
有限元法
COMSOL仿真软件
eddy current magnetic field
conditional generative adversarial neural network(CGAN)
depth conversion
finite element method
COMSOL simulation software
